Succinate receptor GPR91 provides a direct link between high glucose levels and renin release in murine and rabbit kidney (original) (raw)
High glucose level directly triggers renin release in vitro. Direct and acute effects of high glucose levels on the JGA were studied free of systemic influences, using a well established in vitro approach (12, 13) that combines a JGA microperfusion model with fluorescence confocal imaging. Increasing glucose concentration of the afferent arteriole perfusate triggered renin release and vasodilatation of the afferent arteriole within a few minutes (Figure 1 and Supplemental Video 1; supplemental material available online with this article; doi:10.1172/JCI33293DS1). Within 30 minutes of high levels of glucose application, JGA granular content was reduced by 54 ± 5Δ% (where Δ% means the change in quinacrine fluorescence intensity [F0–F] normalized to the baseline [F0]) compared with control basal release (11 ± 5Δ%), and the afferent arteriole dilated significantly. These effects of high glucose level were endothelium dependent, since removing the endothelium of the afferent arteriole (14 ± 1Δ%) or bath application of high glucose level (6 ± 7Δ%) had no effect on renin release. The use of mannitol to control for the effect of osmolality resulted in only minor renin release (19 ± 4Δ%). The actions of high glucose level most likely involved NO and prostaglandins, since both NO synthase and cyclooxygenase inhibition abolished the effects of application of high levels of glucose (14 ± 6Δ% and 13 ± 2Δ% renin release, respectively). Similar effects were observed on the vascular tone.
Direct and acute effects of high glucose level on the JGA. (A and B) Real-time confocal fluorescence imaging of renin content (quinacrine, green) and vascular diameter (cell membranes are labeled with R-18, red) in the in vitro microperfused terminal afferent arteriole (AA) with attached glomerulus (G) freshly isolated from rabbit kidney. In response to increasing glucose concentration of the afferent arteriole perfusate from 5.5 mM (control, A) to 25.5 mM (high glucose level, B), a significant number of renin granules in JG cells (JG) released their fluorescent content and the afferent arteriole internal diameter (arrows) increased. Scale bar: 20 μm. (C) Normalized reductions in quinacrine fluorescence (as an index of renin release) and increases in afferent arteriole diameter within 30 minutes of high glucose level (HG) application. Blockade of NO synthases (L-NAME; 1 mM) and cyclooxygenases ([I] indomethacin; 50 μM) inhibited the effects of high glucose level, indicating involvement of NO and prostaglandins, respectively. Removing the endothelium (endo) or bath glucose had no effect, and equimolar mannitol caused only minor renin release. *P < 0.01; n = 6 each.
Involvement of the TCA cycle intermediate succinate. Since it was established earlier that succinate causes renin release (11), we tested whether the mechanism of the high-level-of-glucose effect involves the TCA cycle. The effects of various TCA cycle inhibitors and intermediates were studied using the same experimental model as above. The strategy is summarized in Figure 2A, and the findings are shown in Figure 2, B–D. First, the addition of 5 mM succinate to the arteriolar perfusate triggered a robust renin release (73 ± 2Δ%), confirming the earlier data (11). The effects of succinate and high glucose levels did not appear to be additive (56 ± 3Δ%; Figure 2B). Then malonate, an inhibitor of the TCA cycle at the succinate dehydrogenase step (succinate degradation), was used. Malonate alone, in the presence of normal glucose (5.5 mM), caused significant renin release (54 ± 7Δ%). Combining malonate with a high glucose level produced an augmented response (77 ± 2Δ%). The effect of malonate was additive to that of succinate as well (83 ± 4Δ%; Figure 2B). Fluorocitrate was next used to block the TCA cycle enzyme aconitase, which is essential for producing succinate. The addition of fluorocitrate alone had no effect on renin release (11 ± 2Δ%), but it abolished the effects of high glucose level on renin release (16 ± 2Δ%). In subsequent studies, the dose-response relationship among succinate or malonate and renin release was established (Figure 2C). The EC50 values for succinate and malonate were 335 and 367 μM, respectively. Administration of α-ketoglutarate (5 mM), another intermediate between the citrate and succinate steps, had no effect (data not shown). Thus, Krebs-cycle intervention studies revealed that renin release specifically occurred in the presence of succinate. Using real-time imaging, the time course of high glucose level– and succinate-induced renin release was established (Figure 2D). Succinate produced a more robust and rapid effect, consistent with the hypothesis that high glucose levels act through accumulating succinate levels. In addition to renin release, we observed substantial vasodilatation of the afferent arteriole within 3–4 minutes after treatment with high glucose levels.
Effects of TCA cycle intermediates and inhibitors on renin release. (A) Overview of the TCA cycle and sites of inhibition with fluorocitrate (F; 100 μM; blocks aconitase in step 2) and malonate (MAL; 1 mM; blocks succinate dehydrogenase in step 6). (B) Effects of TCA cycle inhibitors on renin release in presence of normal (5.5 mM when not indicated) or high glucose (25.5 mM), or succinate (SUCC, 5 mM) levels. C, control. *P < 0.001, control vs. high glucose, high glucose plus succinate, succinate, and malonate. **P < 0.001 malonate alone vs. malonate plus high glucose, and malonate plus succinate. n = 6 each. (C) Dose-dependent effects of succinate and malonate on in vitro renin release. n = 6 measurements for each dose. (D) Representative recordings of changes in quinacrine fluorescence intensity (renin release) in control and in response to high glucose and succinate levels (black lines), and the time course of the high glucose level–induced afferent arteriole vasodilatation (gray line).
Succinate levels in normal and diabetic kidney tissue and urine. To demonstrate local accumulation of succinate in the diabetic kidney tissue to levels that are consistent with GPR91 activation, we measured succinate in nondiabetic and diabetic kidney tissue and urine (Figure 3). In freshly harvested urine and whole kidney tissue samples of nondiabetic mice, the succinate concentration was 26 ± 7.0 and 10 ± 0.2 μM, respectively. In contrast, 1–2 orders of magnitude higher levels were detected in samples from diabetic mice 1 week after streptozotocin (STZ) injection (168 ± 45 μM in urine; 616 ± 62 μM in tissue).
Succinate accumulation in diabetic kidney tissue and urine. An enzyme assay and freshly harvested urine and whole kidney homogenates were used to estimate succinate accumulation in control (n = 5) versus diabetic (DM; 1 week after STZ treatment; n = 6) GPR91 WT mouse kidneys. Individual samples were measured in triplicates and averaged. *P < 0.001 control vs. diabetic.
GPR91 specificity. Similar microperfusion experiments were performed using preparations dissected from GPR91+/+ and GPR91–/– mouse kidneys. Increasing glucose content of the afferent arteriole perfusate from 5.5 to 25.5 mM greatly stimulated the rate of renin release in GPR91+/+ mice, similar to that observed in rabbits (Figure 1C). Granular content was reduced by 44 ± 3Δ% within 30 minutes (Figure 4). Importantly, high glucose level–induced renin release was diminished in GPR91–/– mice; granular content was reduced by only 16 ± 3Δ%. Interestingly, the magnitude of this response is comparable to what was obtained before using mannitol for the control of osmolality effect (Figure 1C). These data suggest that the effect of high glucose level on renin release has a minor hyperosmotic, but a very significant metabolic component. The succinate receptor GPR91 appears to be involved in the metabolic component, consistent with our main hypothesis that high glucose level triggers renin release through the accumulation of succinate and GPR91 signaling.
High glucose level–induced renin release in GPR91+/+ and GPR91–/– mice. Effects of high glucose level (25.5 mM) on renin release, measured by the reduction in quinacrine fluorescence. Hyperosmotic control using mannitol is also shown (same data as in Figure 1C). C, control baseline renin release. GPR91+/+ (WT, n = 6) and GPR91–/– (KO, n = 4) mice were used. *P < 0.001, control vs. high glucose lumen (rabbit, mouse).
Localization of GPR91. Molecular and functional studies were performed to identify the cell type(s) expressing GPR91 and involved in the generation of the renin release signal. RT-PCR was performed to detect GPR91 mRNA in mouse whole kidney and in cell cultures of various JGA cell types (Figure 5A). Consistent with the role of the vascular endothelium in glucose-induced renin release (Figure 1C), expression of the succinate receptor GPR91 was found on the mRNA level in a recently established glomerular endothelial cell (GENC) line (14). In contrast, primary cultures of VSMCs and JG cells produced no signal (for confirmation of cell phenotypes see Supplemental Figure 1). Whole kidney tissue from GPR91+/+ and GPR91–/– mice served as positive and negative controls. To determine the localization of GPR91 protein in the kidney and in cells of the JGA in particular, immunohistochemistry on rat (Figure 5, B–E), mouse, and rabbit (data not shown) kidney sections was performed using a recently produced, commercially available GPR91 polyclonal antibody. Vascular endothelial cells in the afferent arteriole and glomerulus, but not JG cells, were labeled positive for GPR91 (Figure 5B). Endothelial localization was confirmed by double labeling with the rat endothelial marker RECA-1 (Figure 5C). GPR91 and RECA-1 immunolabeling were localized to the same structures (Figure 5D). This is consistent with the RT-PCR data above (Figure 5A) and with our hypothesis that it is the vascular endothelium, rather than other JGA cells, that is capable of detecting extracellular succinate.
Localization of GPR91 in GENCs. (A) Representative RT-PCR demonstrating the presence/absence of GPR91 in various mouse kidney cell types. C, control mix with no cDNA; WK, whole kidney from WT GPR91+/+ or KO GPR91–/– mice. Same results were obtained in n = 6 different samples each. JGC, renin-producing JG cells. (B–E) Localization of GPR91 protein in rat kidney with immunohistochemistry. (B) Strong GPR91 labeling (red) was found in vascular endothelial cells, in both terminal afferent arteriole (Aff. art.) and glomerulus. One region of the glomerulus (indicated by a rectangle) is magnified in C and D for high-resolution colocalization studies. (C) Double labeling for rat endothelial cell marker RECA-1 (green) identified the endothelium of intraglomerular capillary loops (cl). (D) Overlay of GPR91 (red) and RECA-1 (green) images shows colocalization (yellow) of the 2 proteins in the same structures. DIC background is merged with fluorescence to show glomerular morphology. (E) Negative control using no GPR91 primary antibody. Nuclei are blue. Scale bar: 10 μm (B–E). (F) Elevations in GENC [Ca2+]i levels from baseline (under normal glucose [NG], 5.5 mM) in response to high glucose level (25.5 mM) and TCA cycle inhibitors and intermediates. Succinate, 5 mM; malonate, 1 mM; fluorocitrate, 100 μM. *P < 0.001, compared with NG; n = 9 each. (G) Dose-response relationship of high glucose level– and succinate-induced elevations in GENC Ca2+ levels; n = 6 each. F/F0, fluorescence (F) normalized to baseline (F0). (H) Cell- and GPR91-specificity of the succinate-induced [Ca2+]i response. HEK cells were used for biosensor experiments (see below). *P < 0.001, compared with baseline; n = 6 each. Cells were grown on coverslips to near confluency, loaded with Ca2+ sensitive fluorescent dye fura-2, and [Ca2+]i was measured using a cuvette-based spectrofluorometer (Quantamaster-8; Photon Technology Inc.).
Additional functional studies tested whether GENCs that express GPR91, but not VSMCs and JG cells, produce elevations in cytosolic Ca2+ levels in response to the same TCA cycle inhibitors and intermediates that were used with in vitro renin release measurements (Figure 2B). Figure 5F shows significant elevations in GENC intracellular Ca2+ [Ca2+]i levels in response to high glucose level (605 ± 36 nM), succinate (549 ± 29 nM), and malonate (377 ± 22 nM) compared with [Ca2+]i levels measured in presence of normal glucose (133 ± 14 nM). Fluorocitrate failed to increase GENC [Ca2+]i levels (143 ± 11 nM) (Figure 5F). GENCs produced a substantial, dose-dependent Ca2+ signal when high levels of glucose or succinate were added to the superfusate, with an EC50 value of 11 mM and 69 μM, respectively (Figure 5G). These data are consistent with the central role of succinate and the proposed Ca2+-coupled mechanism of GPR91 signaling (4). In addition, saturating doses of succinate did not cause [Ca2+]i elevations in VSMCs, JG cells, or HEK cells (Figure 5H). Also, the succinate-induced response in GENCs was GPR91-specific, since silencing GPR91 with short, interfering RNAs (siRNA) produced no [Ca2+]i response (Figure 5H). Evidence for effective silencing of GPR91 in cultured GENCs using siRNA can be found in Supplemental Figure 2.
Elements of the paracrine signal transduction cascade. Since endothelial cells are indispensable for high glucose level–induced renin release and are the only cell types in the terminal afferent arteriole to express GPR91, they appear to be key components of generating the renin release signal. Fluorescence imaging studies in the microperfused JGA measured endothelial intracellular Ca2+ ([Ca2+]i) changes and NO production (Figure 6, A–C). Increasing glucose concentration from 5.5 to 25.5 mM or the addition of 5 mM succinate to the arteriolar perfusate caused significant elevations in [Ca2+]i and NO production in the vascular endothelium (Figure 6C). In addition, the effect of high glucose level was GPR91-dependent (Figure 6C). Since prostaglandins (PGE2 and PGI2) are classic mediators of renin release, further experiments tested if endothelial [Ca2+]i elevations trigger not only NO but also PGE2 production and release from GENCs (Figure 6D). As expected, GENCs released PGE2 in response to succinate in a dose-dependent fashion with an EC50 value of 214 μM. These chemical mediators, NO and PGE2, are effective vasodilators and were most likely involved in the high glucose level–induced increase in afferent arteriole diameter observed simultaneously with renin release.
GPR91 signaling in the vascular endothelium. Succinate- and GPR91-induced endothelial cytosolic Ca2+ [Ca2+]i signaling and NO production were studied using the microperfused mouse afferent arteriole–attached glomerulus preparation and confocal fluorescence microscopy with fluo-4/fura red ratiometric Ca2+ (A) and DAF-FM imaging (B), respectively. Arrows point at glomerulus and afferent arteriole endothelial cells in situ. Scale bar: 20 μm. (C) Summary of the high glucose– (25.5 mM) and succinate- (5 mM) induced normalized changes in endothelial [Ca2+]i and NO production in WT GPR91+/+ or KO GPR91–/– kidney tissue. *P < 0.05, WT (+/+) vs. KO (–/–); n = 6 each. (D) Dose-response relationship of succinate-induced elevations in PGE2 production and release from cultured GENCs, measured using a PGE2 biosensor. Specially engineered biosensor cells, HEK293 cells, expressing the Ca2+-coupled PGE2 receptor EP1 were loaded with fluo-4 and positioned next to GENCs in culture. Effects of succinate on GENC PGE2 production were measured based on the biosensor cell Ca2+ signal, since upon PGE2-binding these biosensor cells produce a Ca2+ response detected by fluorescence imaging. The cyclooxygenase inhibitor indomethacin (50 μM) and EP1 receptor blocker SC-51322 (10 μM) in presence of 5-mM succinate both served as negative controls for PGE2 specificity. Normalized changes in HEK293-EP1 biosensor cell fluo-4 intensity are shown and served as an index of PGE2 release. n = 6 for each dose.
GPR91 signaling in the STZ model of type 1 diabetes in vivo. Whole animal studies were performed using GPR91+/+ and GPR91–/– mice to evaluate the in vivo importance of GPR91 in renin signaling in diabetic and nondiabetic control animals using the STZ model of type 1 diabetes (STZ-diabetes). A multiphoton imaging approach (15, 16) was used to directly visualize and quantify JGA renin granular content in the intact, living kidney (Figure 7, A–E). Nondiabetic GPR91–/– mice had reduced JGA renin granular content compared with GPR91+/+ littermates (Figure 7, A, B, and E), although total renal renin content was not altered based on whole kidney immunoblotting (Figure 7, F and G). In GPR91+/+ mice, the STZ-diabetes caused a 3.5-fold increase in total renin (Figure 7, F and G) and a less pronounced but still significant increase in JGA renin granular content (Figure 7, C–E). Importantly, both total and JGA renin were significantly reduced in GPR91–/– littermates with STZ-diabetes (Figure 7, E and G).
Renin granular content in GPR91+/+ and GPR91–/– mice in vivo. (A and B) Representative multiphoton fluorescence images of JGA renin content (green) in nondiabetic (A and B) and diabetic (C and D) GPR91+/+ (A and C) and GPR91–/– (B and D) mouse kidneys. The intravascular space (plasma) was labeled using a 70-kDa dextran rhodamine conjugate (red). Scale bar: 20 μm. (E) Summary of changes in JGA renin content in control and in the STZ model of type 1 diabetes (STZ-diabetic; DM) GPR91+/+ and GPR91–/– mice. JGA renin content was calculated based on measuring the area of quinacrine fluorescence (μm2).*P < 0.05, C vs. DM GPR91+/+; Χ_P_ < 0.05, C GPR91+/+ vs. C GPR91–/–; #P < 0.05, DM GPR91+/+ vs. DM GPR91–/– (n = 4 each). (F) Renin immunoblot (green) using control and STZ-diabetic GPR91+/+ and GPR91–/– mouse whole kidney. Four samples are shown for each group. β-actin (red) served as loading control. (G) Summary of changes in whole kidney renin content based on renin immunoblot densitometry. *P < 0.05 C vs. DM GPR91+/+; #P < 0.05 DM GPR91+/+ vs. DM GPR91–/– (n = 4 each).
As an important hallmark and in vivo parameter of diabetes, we measured prorenin levels in kidney tissue and plasma samples of control and diabetic GPR91+/+ and GPR91–/– mice (Figure 8). Diabetes induced a robust, more than 20-fold increase in kidney tissue and plasma prorenin contents in GPR91+/+ mice. In contrast, the increases in kidney tissue and plasma prorenin were significantly blunted in GPR91–/– mice (Figure 8). These results clearly suggest the involvement of GPR91 in the (patho)physiological control of renal renin and prorenin synthesis and release in vivo.
Changes in whole kidney and serum prorenin content in control and STZ-diabetic GPR91+/+ and GPR91–/– mice. Prorenin was measured in whole kidney homogenates and serum as the difference between renin activity before and after trypsin activation of prorenin measured with a fluorescence renin enzyme essay. Compared to age-matched nondiabetic control animals, prorenin content of both whole kidney tissue and plasma samples increased significantly in diabetic animals. Diabetic GPR91–/– animals showed no change in kidney or serum prorenin content. *P < 0.05, C vs. DM GPR91+/+; #P < 0.05, DM GPR91+/+ vs. DM GPR91–/– (n = 4 each).